Artificial neural networks are nonlinear models whose parameters (weights) are estimated during so called training procedure. Applying the back propagation algorithm (which is the most popular method of supervised learning), the computed output (for the initial, usually randomly chosen weights)...